Restaurant marketing with paid ads: a weekly plan

Restaurant marketing with paid ads has one honest limit: the platforms count impressions, clicks and the conversions their pixel or tag saw, and a restaurant's result is a table that was sat or an order that went out. Meta's help page on location targeting says complete accuracy cannot be guaranteed and that you might see impressions, messages or leads from outside your location settings. Google's location assets can enable store visit conversions, but Google enables them only when an account meets its eligibility criteria, so a single venue should not plan on them.

So decide before the first dollar which results exist in a system: a reservation your booking tool confirms, an online order your store records, a call your phone system logs, a code redeemed at the till. Everything else, the walk-in who saw the ad included, is reach. Reach is not worthless, but it is not a cost per result, and a plan that pretends otherwise ends in a return nobody can check.

This guide is a method rather than a promise: four weeks, each with an objective, a radius, a creative, an offer and the figure to read, drawn from Meta's Business Help Center, Google Ads Help and LocaliQ's 2026 benchmarks. Where a figure appears, its source and year follow; where a lift would be convenient, none is claimed.

Meta's Ads Manager has six objectives, and its help page maps the old ones onto them. Awareness reaches the largest number of people most likely to remember the ad, and the old Store traffic objective now lives there, with Store locations features turned on. Traffic sends people to a destination such as your website, Page or Instagram shop, the page's own example being a flash sale. Leads collects leads through messages, phone calls or sign-ups. Sales finds people likely to purchase, through an ecommerce site or by optimizing for an earlier action such as adding to cart. Engagement and App promotion complete the six.

Store locations features are worth knowing and worth qualifying. Meta's page says they help increase awareness of physical stores and encourage foot traffic, show local information such as address or opening hours, offer Call now or Get directions as the call to action, a store locator or Open in map as the destination, and a dynamic field that pulls the address of the nearest store into the text. They are only available with the Awareness objective, and the page opens with a warning: store Pages are only available to managed Facebook clients. A restaurant group with a Meta account team can use them; an independent venue plans on Traffic, Leads and Sales.

On Google, a Search campaign carries location assets: Google Ads Help says they show the store address, a map or the approximate travel distance, with a call button, on Search, Display, YouTube and Google Maps, with the data sourced from Google Maps and your linked Business Profile. Performance Max can be created with the Local store visits and promotions objective; its store goals optimize for store visits, store sales, contacts or directions, store visits and store sales cannot be chosen together, and ads run across Google Maps, YouTube, Gmail, Google.com, the Business Profile page and the Display Network. Local customer optimization inside those campaigns prioritizes people close to or interested in your area on Maps, Waze and local Search formats, and is incompatible with Merchant Center products and online goals.

• Meta: Traffic or Leads for reservations, Sales for online orders, Awareness for an opening
• Meta Store locations: Awareness only, store Pages for managed clients only
• Google: Search with location assets first; Performance Max store goals when visits and directions are the result

Meta's location targeting reaches people who live in, have recently spent time in or go often to countries, regions, cities, postal codes and, in the US, Comscore markets and congressional districts. It is set at the ad set level under Audience and Locations, where you enter or browse locations and choose a location radius; Locations is an audience control, so Meta does not target beyond them unless Reach more people likely to respond to your ads is selected. The radius figures sit in the Marketing API reference: a pin on an address or coordinates takes a radius from 0.63 to 50 miles, or 1 to 80 kilometers, up to 200 pins per ad set, and a city takes 10 to 50 miles. A restaurant group with store sets can let Meta set an automatic radius from population density, or fix one in miles or kilometers, one store set per country.

Google Ads Help describes the same choice with one hard figure: radius targeting requires at least 1 km around a location, and a small radius may make ads show intermittently or not at all. The page names the two fits for a restaurant, a business that delivers within a selected radius and a local business whose customer base is within a radius around the business. It also separates Presence or interest, which reaches people who have shown an interest in the location, from Presence, for advertisers who only want people in the location; a dine-in venue wants Presence, a delivery kitchen wants it even more.

Dayparting is where a restaurant differs from every other local business: the ad for lunch is worthless at 9 pm. Google lets you create up to 6 ad schedules per day for each campaign, in the account's time zone, with bid adjustments by day and hour; Meta's scheduling page says you control the dates and the times ads are shown and that Meta does not deliver outside the schedule, while a small number of people may see an ad minutes after the window closes because delivery and viewing are not the same moment. Schedule the booking ad to the hours before the service it fills, and read that slot's reservations on their own.

• Meta: a pin radius from 1 km (0.63 miles) to 80 km, up to 200 pins per ad set
• Google: at least 1 km; Presence rather than Presence or interest for a venue
• Google: up to 6 schedule blocks a day per campaign, in the account's time zone
• Meta: delivery stops at the schedule's end; a few people see the ad minutes later

Instagram ads for restaurants are not a separate campaign; they are placements of the same ad set. Meta's placements page lists Instagram Feed, on the home feed in the app and on desktop and mobile web; Instagram profile feed, within public profiles of people 18 and up, placed where Meta's delivery system expects them to perform best; Instagram Explore home, as a tile in the Explore grid; Instagram Stories and Instagram Reels, both in the app and on desktop and mobile web; and Instagram search results, when someone types a keyword in the app. The same ad set serves restaurant Facebook ads in Facebook Feed, Marketplace, Stories, Reels and search results.

Meta recommends Advantage+ placements so its delivery system can make the most of the budget, and says that adding more placements does not increase the cost of the ad, that available placements depend on the objective, and that some are still in development in some countries. For a restaurant the practical consequence is to produce once for the vertical surfaces: a short video of a dish being plated, the room at the hour you want to fill, the box arriving at a door. Stories, Reels and Explore show that footage to people within the radius without a second campaign.

Text does the local work the image cannot: the neighborhood, the hours, the service you are filling, the booking link. An ad that shows a dish and says where and when it is served is a restaurant ad; an ad that shows a dish is a food photograph.

An offer is worth running when a system of yours will record its result. A fixed-price lunch produces lunch reservations; a first-order discount produces orders carrying a code; an event night produces tickets or deposits; an opening produces reach and, once the booking link is live, bookings. The trail table sets each restaurant promotion idea against the platform, the result and the cost measure, and where no system records the result, the row says so rather than inventing one.

Three rules of practice keep the reading clean. One offer per campaign, so the cost per reservation belongs to one promotion. One code per platform, so an order carrying it says where the click came from. One slot per daypart, so the early-evening menu is judged on first-seating bookings and not on the whole evening. None of these rules raises a result on its own; they make the result readable.

A bar program adds a platform rule. Meta's Advertising Standards count as alcohol ads the venues and events whose primary business model depends on alcohol sales, invitations to an alcohol venue and the availability of alcohol sales; at a minimum such ads may not be targeted to people under 18, with 21 in the United States, 19 in Canada and 25 in Sweden, and they are not allowed at all in a list of countries that includes Norway, Lithuania, Turkey, Thailand and Russia. Google's alcohol policy allows ads promoting the sale of alcohol and branding ads in permitted countries, and forbids ads that target or appeal primarily to people under the legal purchase age, encourage excessive consumption or imply health benefits. A happy-hour ad is an alcohol ad; set the age targeting before the creative.

• One offer per campaign, one code per platform, one slot per daypart
• Happy hour and bar nights: alcohol ads on Meta, age targeting required
• Opening week buys reach; the booking link turns it into a measurable result

The platforms learn from the conversions they see, and a restaurant's conversions happen where their pixel and tag rarely look: in a booking widget on another domain, in an ordering system, on a phone. Sending the reservation or the order back as a conversion is what lets a Leads or Sales campaign optimize for bookings rather than clicks, and it is only possible when a server of yours knows the booking happened.

Adrails does that part. Its script on your site keeps the visit's click ids for 90 days and records page views that stay in Adrails; your server posts the reservation or order, signed, and Adrails credits it to the campaign and ad that brought the visit and sends it to the Meta dataset you chose and, once your workspace can connect it, to a Google Ads conversion action, within a value limit and a daily limit. Shopify orders arrive on their own through webhooks; a booking that only OpenTable, Resy or a delivery marketplace knows about never reaches Adrails, because Adrails reads no reservation platform. The restaurants use case on this site walks through the setup, the templates per location and the owner's report link.

LocaliQ's 2026 Facebook report gives restaurants a $0.45 cost per click and a 2.68% click-through rate on the traffic objective, no restaurant row for the leads objective and no CPM. Multiply the two and the implied CPM is $0.45 × 2.68% × 1,000, or $12.06 per thousand impressions. Databox's October 2026 median CPM across 2,800 companies and every objective is $7.26, with the middle half between $3.67 and $13.03, and Gupta Media's September 2025 average is $8.37, so a restaurant buying clicks sits in the upper half of what Meta advertisers pay per thousand, which is what a small radius and a traffic objective would predict.

The arithmetic for one venue follows. $300 a week on Meta at $0.45 a click is about 667 clicks, and at a 2.68% click-through rate about 24,900 impressions inside the radius: reach, not bookings. $300 a week on Search at the 2026 Restaurants & Food cost per click of $2.05 is about 146 clicks, and at the row's 8.05% conversion rate about 12 leads; the report's own cost per lead for the row is $30.57, so a location that wants 40 reservation requests a week from Search budgets about $1,223. $300 on each platform is $600 a week, $2,600 a month per location, and $13,000 a month for five locations before creative.

Two cautions. These are averages of other people's accounts, and a benchmark CPC says nothing about how many of the clicks booked; replace each input with your own account's figure as soon as a month of data exists. And December is the most expensive month for Facebook and Instagram ads in Gupta Media's data, with holiday increases of as much as 66%, which lands on the one month a restaurant most wants to be seen: plan the gift card budget on a December CPM, not an annual one.

Read each location on Monday in the same order. Spend by location, against the plan. Impressions and CPM within the radius on Meta, against the location's own previous weeks rather than against a published figure, which changes yearly while the account changes weekly. CTR and CPC against the 2026 restaurant rows, 2.68% and $0.45 on Meta's traffic objective, 6.83% and $2.05 on Search, to see whether the creative and the keywords are in range. Reservations and orders from your own systems, by daypart where the ad was scheduled. Cost per reservation and per order, last.

Change one thing a week. If CPM rose and CTR held, the auction moved: a holiday, a new competitor in the radius, a wider radius than last week. If CTR fell with CPM flat, the creative or the offer tired. If both held and reservations fell, the booking page or the slot is the problem, not the ad. Widening the radius and changing the offer in the same week leaves no way to know which one worked.

Once a quarter, compare with the published benchmarks and with the same weeks last year, because the seasonal pattern repeats and a December rise is not a problem to fix. The rest of the time, the figure that decides the next week's budget is the cost per reservation or per order at each location, read from a system that recorded it.

• Spend, impressions and CPM, CTR and CPC, reservations and orders, cost per result
• One change a week per location
• Benchmarks once a quarter; your own prior weeks every Monday